Utilizing Avidity To Improve Antifreeze Protein Activity: A Type III

Nov 5, 2013 - Antifreeze proteins (AFPs) are ice growth inhibitors that allow the survival of several species living at temperatures colder than the f...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/biochemistry

Utilizing Avidity To Improve Antifreeze Protein Activity: A Type III Antifreeze Protein Trimer Exhibits Increased Thermal Hysteresis Activity Ö zge Can†,‡ and Nolan B. Holland*,† †

Department of Chemical and Biomedical Engineering, Cleveland State University, 2121 Euclid Avenue, Cleveland, Ohio 44115, United States ‡ Department of Medical Biochemistry, School of Medicine, Acibadem University, Kayisdagi Cad., No:32, Atasehir, Istanbul, Turkey ABSTRACT: Antifreeze proteins (AFPs) are ice growth inhibitors that allow the survival of several species living at temperatures colder than the freezing point of their bodily fluids. AFP activity is commonly defined in terms of thermal hysteresis, which is the difference observed for the solution freezing and melting temperatures. Increasing the thermal hysteresis activity of these proteins, particularly at low concentrations, is of great interest because of their wide range of potential applications. In this study, we have designed and expressed one-, two-, and three-domain antifreeze proteins to improve thermal hysteresis activity through increased binding avidity. The three-domain type III AFP yielded significantly greater activity than the one- and two-domain proteins, reaching a thermal hysteresis of >1.6 °C at a concentration of 90% folding comparing the protein band between the 24 and 31 kDa markers (trimer) and the band below the 12 kDa marker (monomer). In the sample that was heated to 50 °C, the trimer:monomer ratio was reduced to approximately 50%, and for the 60 °C sample, nearly all of the trimer was disrupted. For the samples that were heated above 60 °C, only monomer was observed. These results are consistent with the reported instability of the foldon domain at elevated temperatures.25 CD spectra for RD3Nfoldon also confirmed the formation of the foldon trimer. The foldon domain, when it is correctly folded, has a fingerprint at 228 nm,25 and RD3Nfoldon exhibits a pronounced maximum at this wavelength, showing that this protein is correctly folded. Thermal Hysteresis Activity and Ice Crystal Morphology. The RD3N thermal hysteresis data were consistent with previously reported values.18 The three-domain RD3Nfoldon showed significantly increased activity compared to that of the one-domain RD3N or the two-domain RD3Ndimer (Figure 5). This increase was more pronounced for the low-concentration region; e.g., the RD3foldon is approximately 10 times as active as the monomer at 150 μM, whereas at 1.5 mM, the activities are different by a factor of ∼2. In the presence of RD3Nfoldon, small (∼10 μm) hexagonal bipyramidal ice crystals, characteristic of type III AFPs,34 formed (Figure 6). The addition of the foldon domain did not affect the ice crystal morphology. This is not unexpected

where a, b, and c are the coefficients of the second, third, and the last terms of eq 4, respectively.



RESULTS Purification of One- and Two-Domain AFPs. The oneand two-domain AFPs were prepared as a monomer and dimer of the same protein, respectively, the N-terminal domain of RD3 (RD3N). The protein was purified by running the purified RD3N sample through a size exclusion HPLC column under partially reducing conditions (Figure 3). The first peak in Figure 3 corresponds to the dimer followed by the monomer peak. Approximately two-thirds of the sample formed a dimer as indicated by the relative peak sizes. This method allowed proper separation of the monomer and dimer as indicated by SDS−PAGE, which shows that RD3foldon and RD3Ndimer have been isolated (Figure 4). For this gel, the samples were not boiled prior to injection to retain folding of the foldon domain. Characterization of the Three-Domain AFP. The threedomain AFP (RD3Nfoldon) results from the trimerization of the protein consisting of RD3N and the 27-amino acid foldon sequence from bacteriophage T4 fibritin connected by a nineresidue linker. To verify the trimerization of purified RD3Nfoldon, Tris-glycine gels were run using purified samples that were heated to different temperatures prior to being loaded.33 The RD3Nfoldon sample that was injected without 8748

dx.doi.org/10.1021/bi401345b | Biochemistry 2013, 52, 8745−8752

Biochemistry

Article

have been proposed,22,23 but we do not assume a priori any particular scaling relationship between surface coverage and thermal hysteresis activity. We do assume the thermal hysteresis is in some manner dependent on the total surface coverage no matter whether the protein has one, two, or three domains. As will be shown, the three independent data sets allow us to determine the scaling relationship between the total surface coverage and thermal hysteresis activity for this type III AFP. The form of the relationship between the thermal hysteresis activity (TH) and surface coverage is modeled as TH = Aθ n

where the coefficient A and the exponent n are empirically determined. This results in eight parameters needed to fit our data: six equilibrium constants total for the three proteins and the coefficient and exponent from eq 10. Recognizing that the various equilibrium constants can be related to each other allows us to reduce the number of fit parameters. As has been the previously described in the development of the two-domain adsorption model,24 the initial adsorption to the surface can be considered diffusion-limited or adsorption (or collision)limited. Briefly, the adsorption process can be considered a two-step process: bulk diffusion, which brings the protein close to the ice surface, and binding to the surface. If diffusion is significantly faster than binding, the number of domains that can bind will be important because it will increase the number of collisions leading to an increase in the rate of binding. In this case, the equilibrium binding constant will be proportional to the number of domains available for binding. However, if diffusion is significantly slower than binding, an increased number of domains will not be significant because the binding is not the rate-limiting step. Therefore, the equilibrium binding constant will be independent of domain number. Of course, some intermediate relationship could exist if the rates are comparable, but the values should lie between these two extremes. If we assume a diffusion- or collision-limiting case, the fit is reduced by two degrees of freedom by relating the initial binding constants (K1) for the three constructs. When one goes from state I to state II for the two- or threedomain protein, the equilibrium constants are also related. Because the length of the linker that tethers the unbound domains to the surface in each case is the same, it is expected that the equilibrium binding constants for each domain will be the same; however, because the three-domain protein will have two free domains while the two-domain protein will have one domain, the K2 for RD3Nfoldon will be twice as large as that for RD3Ndimer. This removes one more degree of freedom, leaving five fitting parameters for either the diffusion- or collision-limited case or seven parameters for a free fit without assuming diffusion or collision limiting conditions. Fits of the data were performed by minimizing the sum squared errors between the measured and calculated thermal hysteresis data as a function of concentration. The seven parameters were calculated using the constraints described above, as well as for diffusion- and collision-limited cases (Table 1). In each case, the data are fit reasonably well, but the collision-limited case resulted in a slightly smaller sum squared error, indicating a better fit. The collision-limited fit was identical to the fit that was done without assuming a diffusion or collision model, but constraining K1 values to be anywhere between the collision and diffusion conditions (free fit). This supports the idea that the collision relationship is best fit to the

Figure 5. Thermal hysteresis activity for RD3N (green), RD3Ndimer (blue), and RD3Nfoldon (red) as a function of domain concentration measured at a rate of 0.01 °C/min. The data clearly demonstrate that even on a per domain basis the thermal hysteresis activity for the multidomain AFPs is significantly greater at low temperatures.

Figure 6. Ice crystal morphologies in the presence of RD3Nfoldon (0.6 mM). Within the thermal hysteresis window, just prior to the burst (or freezing) point (A), the ice crystal is a well-formed hexagonal bipyramid. Immediately after the burst point (B) and shortly after this (C), the crystal is observed to rapidly grow along the c axis, which is parallel to the plane of the page.

because the match of the ice biding face of the protein to the ice crystal lattice determines the morphology of the ice crystal. Rapid ice growth along the c axis was observed at the freezing point. Application of the Modified Langmuir Isotherm. Measured thermal hysteresis data were fit to appropriate Langmuir or modified Langmuir equations to determine equilibrium binding constants and to gain insight into the surface coverage necessary to generate thermal hysteresis. The standard Langmuir equation θ=

KC 1 + KC

(8)

was used for RD3N, and the previously reported Langmuir equation modified for adsorption of a two-domain protein24 1 1 1 θT = + +1 2K1K 2 C 2K 2 −

2 ⎛ 1 1 1 ⎞ 1 1 + ⎟ + ⎜ 2K 2 ⎠ K1K 2 C ⎝ 2K1K 2 C

(10)

(9)

was used for RD3Ndimer. Equation 5, as derived above, was used for three-domain RD3Nfoldon. These three equations relate the surface coverage (θ) to the solution protein concentration (C) through the equilibrium binding constants (Kn). Our measured data are the thermal hysteresis activity as a function of concentration, so a relationship between θ and thermal hysteresis activity needs to be determined to fit our data. Different scaling relationships 8749

dx.doi.org/10.1021/bi401345b | Biochemistry 2013, 52, 8745−8752

Biochemistry

Article

Table 1. Fit Parameters for Thermal Hysteresis Dataa freeb TH ∝ θ

n

diffusionc

TH ∝ θ

1/2

collisiond

TH ∝ θ

TH ∝ θn

TH ∝ θ

n

proteins with more than one domain exhibited significantly higher thermal hysteresis activity on a per domain basis than the single-domain protein. However, there was little difference among the two-, three-, and four-domain proteins.35 This is in contrast to the results that we observed for the three-domain RD3Nfoldon in this study. As seen for the previously reported two-domain proteins, there is a significant increase in activity of RD3Ndimer versus that of RD3N, but an even greater increase in activity is observed between RD3Ndimer and the three-domain RD3Nfoldon. It was demonstrated with RD3 that the linker is flexible, allowing the two domains to independently bind to the ice surface.36,37 This is what differentiates the three domains of RD3Nfoldon, which are each free to bind independently, whereas in the tandem repeat structure of Nishimiya et al., the binding of the middle domain(s) is restricted by the two neighboring domains.35 On the basis of the fits, the collision-limited adsorption provides the best fit to the data. This is in contrast to the previously reported fit for a dimer and monomer24 based on the HPLC-12 type III AFP.38 The fit to the same equations suggested that the fit was better fit by diffusion-limited adsorption. This may be because the binding affinity of HPLC-12 is significantly greater than that of RD3N as indicated by their equilibrium binding constants, 1.9 and 0.38 mM−1, respectively. Because of the weaker binding of RD3N, it requires a relatively larger number of collisions to result in a binding event, pushing it toward a collision-limited process. It is notable that even though the RD3N monomer has significantly weaker binding than HPLC-12, RD3Nfoldon exhibits greater thermal hysteresis activity than the dimer of HPLC-12, indicating the benefit of the additional domain. The fit of our data suggests a proportional relationship between surface coverage and thermal hysteresis activity. The proposed relationship that thermal hysteresis activity is proportional to the square root of surface coverage23 does not satisfactorily fit our data. The previously reported fit to HPLC-12 also resulted in a similar linear relationship between surface coverage and thermal hysteresis activity.24 Extrapolation of the linear relationship to full surface coverage gives a limit of thermal hysteresis activity for a given AFP domain. In the case of RD3N, the limit is 2.1 °C. As the data reach a surface coverage of only 0.75, it is possible that this relationship breaks down at higher surface coverages.

RD3N K (mM−1) RD3Ndimer K1 (mM−1) K2 RD3Nfoldon K1 (mM−1) K2 K3 TH = Aθn A (°C) n rms error (°C2) Ceff (mM)

0.38

0.053

0.36

0.38

0.35

0.76

0.11

0.36

0.76

0.70

1.4

0.76

4.0

1.4

1.4

1.1

0.16

0.36

1.1

1.1

2.8 21

1.5 27

8.0 22

2.8 21

2.7 21

2.11 1.04 0.021 3.7

2.26 0.5 0.038 14.1

2.15 1.02 0.022 11.2

2.11 1.04 0.021 3.7

2.11 1.0 0.021 3.9

Fit constraint: K2,foldon = 2K2,dimer. bFree: K ≥ K1,dimer ≥ 2K, and K ≥ K1,foldon ≥ 3K. cDiffusion: K = K1,dimer = K1,foldon. dCollision: K = 1 /2K1,dimer = 1/3K1,foldon. a

data. In all three cases, the exponent n, indicating the scaling of TH with surface coverage (θ), is nearly 1. Because a reported model has suggested that TH is proportional to θ1/2,23 we fit the data with the additional constraint of this relationship. This resulted in a significantly larger root-mean-square (rms) error in the fit as well as a deviation from the shape of the data for the one- and two-domain proteins (Figure 7).



DISCUSSION Our data clearly demonstrate that the thermal hysteresis activity of an antifreeze protein can be dramatically increased by increasing the number of active AFP domains available for binding. This is particularly evident at low concentrations. Such increased activity has been shown previously by Nishimiya et al., who showed that multimers of type III AFP showed improved thermal hysteresis activity. They produced multimers of RD3N by tandem repetition of these molecules connected by the original nine-residue linker of the naturally occurring two-domain RD3 protein.35 Their results showed that there was little effect in adding more than two domains. All of the

Figure 7. (A) Fits of thermal hysteresis data for RD3N (green), RD3Ndimer (blue), and RD3Nfoldon (red) as a function of protein concentration. The solid lines represent fits of the data where the thermal hysteresis activity is proportional to surface coverage (θ). The dashed line is a fit based on the thermal hysteresis activity being proportional to θ1/2. (B) Thermal hysteresis activity as a function of surface coverage based on the collisionlimited model. This illustrates the linear relationship between thermal hysteresis activity and surface coverage. 8750

dx.doi.org/10.1021/bi401345b | Biochemistry 2013, 52, 8745−8752

Biochemistry



The equilibrium binding constant for the third domain of RD3Nfoldon, as fit to the data, is nearly 1 order of magnitude larger compared to the binding constant of the second domain. This would not be expected if the unbound domain were completely free to diffuse at the end of its linker; however, it could occur if the domain were forced into an orientation that promotes its binding to the surface. This is possible because the three linkers come out one direction from the foldon domain, so once two domains are bound, the third is directed to the surface. When there are two free domains, the domains have greater freedom of motion. The increased level of binding of the third domain leads to a higher surface coverage than what would occur if the third domain were not oriented, directly contributing to the enhanced activity. The successful fit of our data by the Langmuir and modified Langmuir models suggests that these models are adequate for describing the behavior of the system. Although other approaches such as statistical thermodynamics may provide a slightly more accurate fit to the data,39 they would not significantly impact the essence of our results.

REFERENCES

(1) Jia, Z. C., and Davies, P. L. (2002) Antifreeze proteins: An unusual receptor-ligand interaction. Trends Biochem. Sci. 27, 101−106. (2) Young, H. M., and Fletcher, G. L. (2008) Antifreeze protein gene expression in winter flounder pre-hatch embryos: Implications for cryopreservation. Cryobiology 57, 84−90. (3) Chao, H. M., Davies, P. L., and Carpenter, J. F. (1996) Effects of antifreeze proteins on red blood cell survival during cryopreservation. J. Exp. Biol. 199, 2071−2076. (4) Amir, G., Rubinsky, B., Kassif, Y., Horowitz, L., Smolinsky, A. K., and Lavee, J. (2003) Preservation of myocyte structure and mitochondrial integrity in subzero cryopreservation of mammalian hearts for transplantation using antifreeze proteins: An electron microscopy study. European Journal of Cardio-Thoracic Surgery 24, 292−297. (5) Soltys, K. A., Batta, A. K., and Koneru, B. (2001) Successful nonfreezing, subzero preservation of rat liver with 2,3-butanediol and type I antifreeze protein. J. Surg. Res. 96, 30−34. (6) Inada, T., and Modak, P. R. (2006) Growth control of ice crystals by poly(vinyl alcohol) and antifreeze protein in ice slurries. Chem. Eng. Sci. 61, 3149−3158. (7) Egolf, P. W., and Kauffeld, M. (2005) From physical properties of ice slurries to industrial ice slurry applications. Int. J. Refrig. 28, 4−12. (8) Knight, C. A., Cheng, C. C., and Devries, A. L. (1991) Adsorption of α-helical antifreeze peptides on specific ice crystalsurface planes. Biophys. J. 59, 409−418. (9) Kristiansen, E., and Zachariassen, K. E. (2005) The mechanism by which fish antifreeze proteins cause thermal hysteresis. Cryobiology 51, 262−280. (10) Pertaya, N., Marshall, C. B., DiPrinzio, C. L., Wilen, L., Thomson, E. S., Wettlaufer, J. S., Davies, P. L., and Braslavsky, I. (2007) Fluorescence microscopy evidence for quasi-permanent attachment of antifreeze proteins to ice surfaces. Biophys. J. 92, 3663−3673. (11) Celik, Y., Drori, R., Pertaya-Braun, N., Altan, A., Barton, T., BarDolev, M., Groisman, A., Davies, P. L., and Braslavsky, I. (2013) Microfluidic experiments reveal that antifreeze proteins bound to ice crystals suffice to prevent their growth. Proc. Natl. Acad. Sci. U.S.A. 110, 1309−1314. (12) Chao, H., Hodges, R. S., Kay, C. M., Gauthier, S. Y., and Davies, P. L. (1996) A natural variant of type I antifreeze protein with four icebinding repeats is a particularly potent antifreeze. Protein Sci. 5, 1150− 1156. (13) Leinala, E. K., Davies, P. L., Doucet, D., Tyshenko, M. G., Walker, V. K., and Jia, Z. C. (2002) A β-helical antifreeze protein isoform with increased activity: Structural and functional insights. J. Biol. Chem. 277, 33349−33352. (14) Gauthier, S. Y., Marshall, C. B., Fletcher, G. L., and Davies, P. L. (2005) Hyperactive antifreeze protein in flounder species. FEBS J. 272, 4439−4449. (15) Marshall, C. B., Chakrabartty, A., and Davies, P. L. (2005) Hyperactive antifreeze protein from winter flounder is a very long rodlike dimer of α-helices. J. Biol. Chem. 280, 17920−17929. (16) Sonnichsen, F. D., Sykes, B. D., Chao, H., and Davies, P. L. (1993) The nonhelical structure of antifreeze protein type-III. Science 259, 1154−1157. (17) Marshall, C. B., Daley, M. E., Sykes, B. D., and Davies, P. L. (2004) Enhancing the activity of a β-helical antifreeze protein by the engineered addition of coils. Biochemistry 43, 11637−11646. (18) Miura, K., Ohgiya, S., Hoshino, T., Nemoto, N., Suetake, T., Miura, A., Spyracopoulos, L., Kondo, H., and Tsuda, S. (2001) NMR analysis of type III antifreeze protein intramolecular dimer: Structural basis for enhanced activity. J. Biol. Chem. 276, 1304−1310. (19) Esser-Kahn, A. P., Trang, V., and Francis, M. B. (2010) Incorporation of antifreeze proteins into polymer coatings using siteselective bioconjugation. J. Am. Chem. Soc. 132, 13264−13269. (20) Can, O., and Holland, N. B. (2011) Conjugation of type I antifreeze protein to polyallylamine increases thermal hysteresis activity. Bioconjugate Chem. 22, 2166−2171.



CONCLUSIONS We have successfully expressed and characterized type III AFP constructs to increase thermal hysteresis activity. Foldon, an oligomerization domain, was used to create a type III AFP trimer that is more active than the one-domain type III AFP, yielding a 1.65 °C thermal hysteresis activity at 0.6 mM. In addition, we have developed a new model to describe threedomain protein adsorption based on the classical Langmuir model. Using classical and modified Langmuir models, the relationship between surface coverage and thermal hysteresis activity and equilibrium binding constants were obtained for a type III antifreeze protein. It should be noted that modified Langmuir isotherms derived in this study can be of general use and can be further modified and extended to other molecular constructs with more than three independent binding regions. The increased activity is attributed to avidity, which increases activity at low concentrations but does not increase the maximal theoretical thermal hysteresis activity. To accomplish this, the oligomerization approach could be extended to hyperactive insect AFPs to increase the maximal thermal hysteresis activity. Nevertheless, designing AFP constructs that have higher thermal hysteresis activities at significantly lower concentrations would make it much more economical to use them as cryopreservation additives and ice slurry stabilizers.



Article

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: (216) 687-2572. Funding

This research was supported by the American Heart Association (Scientist Development Grant 0635084N) and the Cleveland State University Doctoral Dissertation Research Expense Award Program. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank A. H. Heuer and A. McIlwain (Case Western Reserve University) for the use of and assistance with the nanoliter osmometer. 8751

dx.doi.org/10.1021/bi401345b | Biochemistry 2013, 52, 8745−8752

Biochemistry

Article

(21) Graham, L. A., Marshall, C. B., Lin, F. H., Campbell, R. L., and Davies, P. L. (2008) Hyperactive antifreeze protein from fish contains multiple ice-binding sites. Biochemistry 47, 2051−2063. (22) Burcham, T. S., Osuga, D. T., Yeh, Y., and Feeney, R. E. (1986) A kinetic description of antifreeze glycoprotein activity. J. Biol. Chem. 261, 6390−6397. (23) Wang, S., Amornwittawat, N., and Wen, X. (2012) Thermodynamic analysis of thermal hysteresis: Mechanistic insights into biological antifreezes. J. Chem. Thermodyn. 53, 125−130. (24) Can, O., and Holland, N. B. (2009) Modified Langmuir isotherm for a two-domain adsorbate: Derivation and application to antifreeze proteins. J. Colloid Interface Sci. 329, 24−30. (25) Frank, S., Kammerer, R. A., Mechling, D., Schulthess, T., Landwehr, R., Bann, J., Guo, Y., Lustig, A., Bachinger, H. P., and Engel, J. (2001) Stabilization of short collagen-like triple helices by protein engineering. J. Mol. Biol. 308, 1081−1089. (26) Stetefeld, J., Frank, S., Jenny, M., Schulthess, T., Kammerer, R. A., Boudko, S., Landwehr, R., Okuyama, K., and Engel, J. (2003) Collagen stabilization at atomic level: Crystal structure of designed (GlyProPro)(10)foldon. Structure 11, 339−346. (27) Ghoorchian, A., Vandemark, K., Freeman, K., Kambow, S., Holland, N. B., and Streletzky, K. A. (2013) Size and shape characterization of thermoreversible micelles of three-armed star elastin-like polypeptides. J. Phys. Chem. B 117, 8865−8874. (28) Guex, N., and Peitsch, M. C. (1997) SWISS-MODEL and the Swiss-PdbViewer: An environment for comparative protein modeling. Electrophoresis 18, 2714−2723. (29) Gill, S. C., and von Hippel, P. H. (1989) Calculation of protein extinction coefficients from amino-acid sequence data. Anal. Biochem. 182, 319−326. (30) Chakrabartty, A., and Hew, C. L. (1991) The effect of enhanced α-helicity on the activity of a winter flounder antifreeze polypeptide. Eur. J. Biochem. 202, 1057−1063. (31) Wang, X., DeVries, A. L., and Cheng, C. C. (1995) Antifreeze peptide heterogeneity in an antarctic eel pout includes an unusually large major variant comprised of two 7 kDa type III AFPs linked in tandem. Biochim. Biophys. Acta 1247, 163−172. (32) Ghoorchian, A., and Holland, N. B. (2011) Molecular architecture influences the thermally induced aggregation behavior of elastin-like polypeptides. Biomacromolecules 12, 4022−4029. (33) Ghoorchian, A., Cole, J. T., and Holland, N. B. (2010) Thermoreversible micelle formation using a three-armed star elastinlike polypeptide. Macromolecules 43, 4340−4345. (34) Davies, P. L., Baardsnes, J., Kuiper, M. J., and Walker, V. K. (2002) Structure and function of antifreeze proteins. Philos. Trans. R. Soc., B 357, 927−933. (35) Nishimiya, Y., Ohgiya, S., and Tsuda, S. (2003) Artificial multimers of the type III antifreeze protein: Effects on thermal hysteresis and ice crystal morphology. J. Biol. Chem. 278, 32307− 32312. (36) Holland, N. B., Nishimiya, Y., Tsuda, S., and Sonnichsen, F. D. (2007) Activity of a two-domain antifreeze protein is not dependent on linker sequence. Biophys. J. 92, 541−546. (37) Holland, N. B., Nishimiya, Y., Tsuda, S., and Sönnichsen, F. D. (2008) Two domains of RD3 antifreeze protein diffuse independently. Biochemistry 47, 5935−5941. (38) Baardsnes, J., Kuiper, M. J., and Davies, P. L. (2003) Antifreeze protein dimer: When two ice-binding faces are better than one. J. Biol. Chem. 278, 38942−38947. (39) Quiroga, E., and Ramirez-Pastor, A. J. (2013) Statistical thermodynamics of molecules with multiple adsorption states: Application to protein adsorption. Chem. Phys. Lett. 556, 330−335.

8752

dx.doi.org/10.1021/bi401345b | Biochemistry 2013, 52, 8745−8752